[07:21:31] gih joins the room [07:29:44] kurtis joins the room [07:46:21] kurtis leaves the room [07:52:51] kurtis joins the room [07:57:21] kurtis leaves the room [08:03:38] kurtis joins the room [08:07:41] Is the audio ok? [08:07:57] I can't find a link to any audio [08:08:10] I have no idea what Joel is up to! [08:08:12] :-( [08:10:19] fujiwara joins the room [08:10:25] shane joins the room [08:10:27] I think they are having some issues [08:10:31] wouter joins the room [08:10:32] swb joins the room [08:10:32] randy joins the room [08:10:33] So I guess its being recorded, but I have no idea if its being streamed out - tghere is not obvious that links me to a styream [08:10:45] gigix73 joins the room [08:11:07] at least listening in on the swedish disucssion in the back [08:11:08] kurtis leaves the room [08:11:19] marc.blanchet.qc joins the room [08:12:03] nope - I really cannot see any links to an audio stream [08:12:19] paf joins the room [08:12:21] modify the host's network stack is not that difficult? [08:13:00] antoin joins the room [08:13:18] Marc: 90% use Windows, 9% use OS X, 1% use Linux (on the desktop)... problem solved with 3 OS! [08:13:27] ;) [08:13:40] jakob joins the room [08:14:02] "Windows" is 5 different OSs [08:14:15] not counting mobile OS... ;-) [08:15:02] kurtis joins the room [08:18:12] kurtis leaves the room: Replaced by new connection [08:18:17] kurtis joins the room [08:20:36] kurtis leaves the room: Replaced by new connection [08:20:40] kurtis joins the room [08:24:01] rhe joins the room [08:24:05] wouter leaves the room [08:25:28] gih leaves the room [08:25:57] wouter joins the room [08:26:41] Jelte joins the room [08:28:37] wouter leaves the room [08:28:37] wouter joins the room [08:29:04] Roy Arends joins the room [08:29:12] I like the notifications in the upper right of paf's screen. [08:30:13] bad combination, a lot of jabber buddies who are all on the same intermittent wireless connection... [08:30:13] Bonus points for getting a message up there... [08:30:23] Tony Li joins the room [08:31:34] wouter leaves the room [08:35:27] double bonus points for getting growl on projected screen but not in this room [08:36:16] Simon joins the room [08:36:19] pawal joins the room [08:36:37] Simon Perreault joins the room [08:36:38] I won! [08:36:45] :-) [08:41:07] ggm joins the room [08:41:15] audiocast? [08:41:21] wouter joins the room [08:41:38] marc.blanchet.qc leaves the room [08:42:20] randy, how about "who wants less TLDs" [08:42:35] It was not only jabber...also twitter and stuff.... [08:42:39] fair question [08:42:51] (so I guess silence here == no audiocast) [08:42:55] I was working on a different presentation, but Kurtis called me a bit too early, and I forgot the growl... :-) [08:43:01] If folks figure out how to monetize TLD's, there'll be demand. [08:43:08] George - yeah Lixia is talking, using the microphone. [08:43:28] she'll have to turn up the volume to get the sound here to Brisbane [08:47:04] oh there is demand already, but i wonder if there are more people who want it than people who really don't [08:47:21] arifumi joins the room [08:50:27] I think demand in cctld space is fine, (even vanity cc's like .me &c) all about the flag, less about the cash. [08:51:13] problem is the cost is not colocated with the benefit. [08:53:28] randy: so you're saying TLD namespace expansion is like spam? [08:54:27] or like interdomain routing? [08:55:07] Simon Perreault leaves the room [08:55:31] I think the problem is that economic rules or calculations don't apply to TLD's, as there is a vector that does not excist in normal econlomic traffic: We don't wnat to go back to the host.txt, and that vector needs to be incorporated in the economic model of question and demand [08:57:01] i both hope and fear that the money from the first 50 tld registrations will be used entirely for hiring lawyers [08:57:19] Only the first 50? [08:57:37] To deter the next 50 [08:57:40] after that there's budget for a few extra servers :) [08:58:02] Jelte, the money is collected by ICANN. Do they buy the servers? [08:58:25] icann would gladly control the servers [08:58:43] notice i said "gladly" not "competently" [08:59:10] april 1st draft potential: with idn's in klingon script and colonization of other planets, should we propose draft-ietf-interplanetary-dns? Think about query response delays, a new interplanetary root (current becomes .earth in all possible scripts). etc, etc. [08:59:22] maybe not the root servers, but if you're expecting a gazillion registrations you better be prepared to handle those [09:00:05] Jelte: Correct, now think about things in your statements like "you" and "prepared", and what they mean. When you have an idea, let us know! Seriously! [09:00:28] wouter leaves the room [09:00:31] ljb joins the room [09:00:41] wouter joins the room [09:01:03] i will :) [09:01:06] Roy: only if the IETF is delegated the .etc domain. [09:01:26] SwedeMike joins the room [09:01:54] You mean wildcard Shane ? [09:01:55] ray joins the room [09:02:23] pawal leaves the room [09:02:25] pawal joins the room [09:02:34] I'd like to register a single character $2A in the root. [09:02:56] wildcard shane? cool nicname [09:02:57] let's take up a collection and register .FOAD [09:03:30] So why didn't you raise your finger if you want the root to expand ? [09:03:32] www.make-money-fast.wildcard\x20shane [09:04:06] shane, that should be: WWW.MAKE-MONEY....... . [09:04:28] make a new registry .tld where all the new 'tlds' can register themselves. [09:05:03] problem there is, under who'se control.. [09:05:18] i nominate wouter [09:05:50] I nominate Bert [09:06:26] I nominate Postel. He can probably _still_ do a better job than ICANN. [09:06:32] ljb leaves the room [09:07:03] Geoff Huston joins the room [09:07:04] ljb joins the room [09:15:00] Jaap Akkerhuis joins the room [09:15:02] How did it look for South America? [09:16:12] It is politically VERY dangerous to use terminology like "country announce addresses", or "address space used in a country" (or even "by a country") [09:16:26] is she using geoip locatation database [09:16:26] brasil has lots, rest was light, but not as bad as the look of africa. [09:16:34] or the RIR allocation country codes [09:16:35] ggm, probably you found this yourself by now, but http://feed.verilan.com/ietf/stream06.m3u [09:16:35] apparently [09:17:16] Geoff: RIR data, I believe. [09:17:17] thank you. [09:17:25] ooops [09:19:07] probably better to use one of the geolocation databases for this kind of study I would've thought [09:19:28] I'd claim that it doesn't matter much [09:19:46] That was just the impression I got from the earlier discussion, I could be wrong, may be worth a clarifying question. [09:19:57] I think we should stop trying to map address space to countries [09:20:05] the US country code would have some international dispersal [09:20:07] What conclusions are being drawn or implied from it? [09:20:11] and RIPE use EU as a country code [09:20:37] We already have requests that IPv6 address space is pre-allocated so that every country get the same number of addresses. "Just like frequency spectrum" [09:21:04] Note, every _country_ get the same number of addresses. [09:21:09] oh wow [09:21:14] if ipv6 really was infinitely large, this could be done ... [09:21:49] So, why are we (who know this is not possible) continuing to talk about Ip address space per country? (I will talk with Lixia of course...I have not done it yet) [09:21:55] pawal leaves the room [09:21:56] I think Lixia is not rally representing RIR allocation practices at all well. [09:22:01] pawal joins the room [09:22:55] Some grad student thought it would make for a pretty picture for a presentation. [09:23:09] If the projector was working properly, it would be pretty. [09:23:23] so these are breakouts, isn't that done mainly in the ARIN region? [09:23:27] Please untwist knickers [09:23:30] paf: in theory, ip addresses are used by machines which have locations. As an analogy, UK has about 6 or 7M registrations under co.uk. Would it be conceivable to determine the number of addresses used in the UK? [09:23:40] ggm leaves the room [09:24:28] yes, Roy, but what does it mean? [09:25:03] ljb leaves the room [09:25:39] Tables like http://resources.potaroo.net/iso3166/v4cc.html are easy to construct, but appear to not really offer very much in the way of detailed insight in network dispersion and behaviours [09:25:39] let's hope we don't have to build networks that are limited by country lines, as that would most likely be suboptimal (I know several cases where traffic from sweden to sweden passes thru denmark or norway (so that's even leaving the EU)) [09:25:48] Roy: The answer I have heard so far (and this is what the story have been) is "No", as the implications of a "yes" answer is that SPs that have a network that cross country boundaries MUST ensure they do not move IP addresses across that boundary without updating some registry, ensuring the use of the address is according to licensing terms in that country etc. [09:25:54] ljb joins the room [09:26:11] SwedMike: Exactly! Same explanation but different words. [09:26:19] Brenden Kuerbis joins the room [09:26:43] i have traceroutes to machines in buildings i can see across the street that go through switzerland :p [09:27:05] paf leaves the room [09:27:52] Routing doesn't follow geography... ;-) [09:28:15] I can make packets from San Jose to San Jose travel through Plano TX. [09:28:32] BFD [09:28:36] there are two things here. 1) I said number of addresses used, not which addresses used. 2) folks seem to think that the only way to determine # addresses is by looking at ISP. [09:32:46] when was Randy last stable? [09:34:33] 1948? [09:34:39] around 90% of the ASes in BGP are stub and not transit [09:35:01] so widespread use of default in stub ASes is not surprising [09:35:44] taking full tables is expensive, so it's not surprising [09:35:49] default is cheap [09:38:54] Jaap Akkerhuis leaves the room [09:39:30] but if 100% of small and medium ISPs were default-free, then the stub:s would be default-free as well because they couldn't reach much :P [09:39:47] so stubs should require their upstreams to be default free [09:39:56] Why? [09:40:11] why what? [09:40:12] I want my upstream to deliver my data... [09:40:43] yes, but do you want them to deliver your data even if it's destined for a prefix not in DFZ? [09:40:59] I do NOT care if it's in the DFZ or not. [09:41:10] I want it to go to the site advertising the prefix. [09:41:26] then I guess we're all in disagreement whether default:ing is a security problem or not [09:41:32] Hopefully the folks that actually own the prefix, and the folks that I think I'm conversing with. [09:42:01] The remarkable thing is that there was a default path back to Randy, across the alleged DFZ. The little ISPs assume they can be casual because the DFZ won't be. [09:42:10] Why is a default a security problem? [09:42:42] shane_: because you can try to secure the BGP control plane but still forward packets to routes rejected by BGP policy [09:43:03] so anyone who can get in anywhere can suck up packets [09:43:08] Only if there are no prefixes accepted by BGP policy [09:43:38] But doesn't having a default imply you will attempt to send packets to anywhere? Isn't that your policy? [09:43:39] This is no worse than your upstream executing an intelligent MITM attack. [09:43:44] wouter leaves the room [09:43:53] ljb leaves the room [09:43:55] wouter joins the room [09:43:58] No, having default means that you depend on upstream. [09:44:06] pawal leaves the room [09:44:07] pawal joins the room [09:44:23] [09:44:30] ljb joins the room [09:44:30] You'll send the packets to your upstream though, right? [09:44:39] kurtis leaves the room [09:44:42] Of course. [09:44:51] and the problem is when the upstreams themselves have a default etc [09:46:02] No, that doesn't create a problem. It's just a recursive application of the same situation. [09:46:19] The problem happens if an upstream accepts a prefix that they shouldn't have. Same for the DFZ. [09:46:58] Roy Arends leaves the room [09:47:13] So the real "security problem" is overzealous network administrators doing a lot of work on their policy and not understanding that setting a default bypasses all of this work, right? [09:47:13] What about the prices of storage BANDWIDTH? [09:47:33] Storage gets bigger for the same price, but it doesn't get faster at the same rate. [09:47:58] UK datacenter transit has dropped 40 fold in 10 years [09:48:23] It was £200 per Mbps in late 90s, and can be had for £5 now. [09:48:23] No, that's not a problem. If you set default, you depend on your upstream. If you don't trust your upstream, then you're fooling yourself. [09:48:39] Still not seeing a problem then. [09:48:56] Roy Arends joins the room [09:48:57] Unfortunately DSL backhaul is still £100 per Mbps [09:49:36] cost of transit in sweden is around 2-10 EUR/meg/month [09:49:49] we havent cache:ed anything for 6-8 years, it just doesnt make sense [09:50:09] SwedeMike, so you don't use Akamai? [09:50:28] content delivery networks is another thing, I don't call that "caching" [09:50:36] ah [09:50:43] it's cooperative, and it's done by the content owner/provider [09:50:59] so akamai makes a lot of sense, but doing caching by ISP doesn't [09:52:12] Roy Arends leaves the room [09:53:16] randy leaves the room [09:53:22] To use this use bulk, low priority data, that you can store on cheap disks to wait for cheap transit times (so it gets sent around late at night). [09:55:03] A 1 TB disks costs £50. To fill it in a month I'd need to do 3.2 Mbps continuously. That's pretty good if you're only costing against transit, but horrible if you're doing that over broadband. [09:55:22] wouter leaves the room [09:55:27] swb leaves the room [09:55:36] Tony Li leaves the room [09:55:46] Brenden Kuerbis leaves the room [09:55:52] jakob leaves the room [09:55:53] antoin leaves the room [09:55:57] ray leaves the room [09:55:57] gigix73 leaves the room [09:55:58] rhe leaves the room [09:55:59] ray_: people here have dark fiber to COs, so "broadband backhaul" is... well... not so expensive because it's 1GE over dark fiber [09:56:01] pawal leaves the room [09:56:05] ljb leaves the room [09:56:07] arifumi leaves the room [09:56:07] Jelte leaves the room [09:56:23] shane leaves the room [09:56:32] Geoff Huston leaves the room [09:57:12] Simon leaves the room [10:01:40] Brenden Kuerbis joins the room [10:02:12] Brenden Kuerbis leaves the room [11:08:09] pawal joins the room [11:12:18] ggm joins the room [11:15:46] pawal leaves the room [11:15:55] arifumi joins the room [11:16:34] arifumi leaves the room [11:32:29] fujiwara leaves the room [12:19:15] ggm leaves the room [12:26:41] Jaap Akkerhuis joins the room [12:54:18] Jaap Akkerhuis leaves the room [14:00:12] Jaap Akkerhuis joins the room [14:21:55] Jaap Akkerhuis leaves the room [17:48:02] Jaap Akkerhuis joins the room [17:48:14] Jaap Akkerhuis leaves the room [17:50:44] Jaap Akkerhuis joins the room [18:28:53] Jaap Akkerhuis leaves the room [22:59:12] ggm joins the room [23:14:13] ggm leaves the room [23:16:19] ggm joins the room